Abstract
Phytoplankton is the basis for aquatic food webs and mirrors the water quality. Conventionally, phytoplankton analysis has been done using time consuming and partly subjective microscopic observations, but next generation sequencing (NGS) technologies provide promising potential for rapid automated examination of environmental samples. Because many phytoplankton species have tough cell walls, methods for cell lysis and DNA or RNA isolation need to be efficient to allow unbiased nucleic acid retrieval. Here, we analyzed how two phytoplankton preservation methods, three commercial DNA extraction kits and their improvements, three RNA extraction methods, and two data analysis procedures affected the results of the NGS analysis. A mock community was pooled from phytoplankton species with variation in nucleus size and cell wall hardness. Although the study showed potential for studying Lugol-preserved sample collections, it demonstrated critical challenges in the DNA-based phytoplankton analysis in overall. The 18S rRNA gene sequencing output was highly affected by the variation in the rRNA gene copy numbers per cell, while sample preservation and nucleic acid extraction methods formed another source of variation. At the top, sequence-specific variation in the data quality introduced unexpected bioinformatics bias when the sliding-window method was used for the quality trimming of the Ion Torrent data. While DNA-based analyses did not correlate with biomasses or cell numbers of the mock community, rRNA-based analyses were less affected by different RNA extraction procedures and had better match with the biomasses, dry weight and carbon contents, and are therefore recommended for quantitative phytoplankton analyses.